tomsing1

#!/usr/bin/env bash
set -e
set -o nounset

# This script uses a GTF file (e.g. generated by Stringtie) and
# 1. retrieves BAM files for individual samples from AWS S3
# 2. quantifies the exons defined in the GTF file with featureCounts,
#    e.g. for differential splicing analysis with limma::diffSplice
# 
# Alignments are expected to be reverse-stranded and paired-end.

tomsing1

# Commented out IPython magic to ensure Python compatibility.
# %%capture
# !pip install awswrangler --quiet

import awswrangler as wr
import pandas as pd
import boto3
from google.colab import data_table, userdata

data_table.enable_dataframe_formatter()

tomsing1

import boto3
from google.colab import userdata

# refresh the temporary credentials via the key symbol in the left hand tab
s3_resource = boto3.resource('s3',
    aws_access_key_id=userdata.get('AWS_ACCESS_KEY_ID'),
    aws_secret_access_key=userdata.get('AWS_SECRET_ACCESS_KEY'),
    aws_session_token=userdata.get('AWS_SESSION_TOKEN')
)

tomsing1

library(glue)
library(RSQLite)

con <- dbConnect(RSQLite::SQLite(), ":memory:")
var = "test"
glue_sql("{var}", .con = con)            # <SQL> 'test'
glue_sql("{`var`}", .con = con)          # <SQL> `test`
glue_sql("{DBI::SQL(var)}", .con = con)  # <SQL> test (unquoted)
glue_sql("`{var}`", .con = con)          # <SQL> `'test'` NOT USEFUL

tomsing1

# The following code is based on the shiny::conditionalPanel() help page. The `ui` and `server`
# components have been compartmentalized into the `mod_histogram` module. The `histogramApp`
# function shows an example of using this module.
# Note that the `ns = ns` argument needs to be passed ot the conditionalPanel() call.

library(shiny)

mod_histogram_ui <- function(id){
    ns <- NS(id)
    fluidPage(

tomsing1

library(glue)
library(httr)
library(jsonlite)
library(xml2)

#' Query ENA's REST API for information about records
#' 
#' @param accessions Character vector of one or more ENA record identifiers
#' @return An `xml_document` object
get_records <- function(accessions) {

tomsing1

library(palmerpenguins)
library(shiny)
library(shinyjqui)

categories <- colnames(penguins)[vapply(penguins, is.factor, logical(1))]

server <- function(input, output) {
  lapply(categories, \(category) {
    output[[category]] <- renderPrint({ print(input[[category]]) })
  })

tomsing1

#' Create an image tag with an example image
#'
#' @param width Scalar integer, the width of the image
#' @param height Scalar integer, the height of the image
#' @param title Scalar character, the title of the image
#' @return A `shiny.tag` with the URL to a random image from
#' [Lorem Picsum](https://picsum.photos/)
#' @export
#' @importFrom htmltool tags
#' @importFrom checkmate assert_count assert_character

tomsing1

There is a lot of confusing information about virtual environments in python out there, in part because the tool chain has evolved over many years.

I decided to follow the advice of Real python and The hitchhiker's guide to python and manage both multiple python versions and multiple virtual environments with pyenv

tomsing1

---
title: "Embedding R into Quarto documents with quarto-webr"
subtitle: "Example: intersecting differential expression results"
author: "Thomas Sandmann"
date: '2023/11/18'
format: html
engine: knitr
webr: 
  show-startup-message: true
  packages: ['ggvenn', 'huxtable']